Research Note 84-42

THE ROLE OF JOB SATISFACTIO14 IN ABSENCE BEHAVIOR

Daniel R. Ilgen and John H. HollenbackPurdue University

0 0 T. Owen Jacobs, Contracting Officer's Representative

0

Submitted by

Robert M. Sasmor, DirectorBASIC RESEARCH

____DTiC

___FEB_ 2 21984

U. S. Army A0= Research Institute for the Behavioral and Social Sfeces

I~l February 1984

Approved for public release; distribution unlimited.

[This report, s submitted by the contractor, he. been cleared for reasen to Defense Technical information Centers(DTIC) to comply wvith regulatory requirements. It has been given no primary distribution other than to OTICOWd will be evailable only through DTIC or other reference services such as the National Technical InformationService INTIS). The vitws, opinlene, and/or findings contained in this report are those of the author(s) andshiould not be construed a an official Depaetmient of the Army position, policy, or decision, unless to desinatjby other oflia douetain 84 02 21 151

SIECURITY CLASSIFICATION OF T141S PAGE (When Data EnteredQ

* REPORT DOCUMENTATION PAGE BEOECMLTGFR1REPORT NUMBER 12. GOVT ACCESSION NO. 3. RECIPIENT'S CATALOG NU IMBER

Research Note 84-424. TITLE (andSubtitle) S. TYPE OF REPORT &PERIODOCOVERED

THRE ROLE OF JOB SATISFACTION IN-ABSENCE BEHAVIOR Interim:-Report.Series 1975-1976

- 6. PERFORMING ORO. REPORT NUMBER

7. AUTMOR(et) S. 'CONTRACT OR GRANT NUMBER(o)

Daniel R. Ilgen and John.H. Hollenback DAHC19-74-G-0002

9. PERFORMING ORGANIZATION NAME AND ADDRESS I0. PROGRAM ELEMENT. PROJECT. TASKAREA & WORK UNIT NUMBERSPurdue Research-Foundation

Division of :Spons-6red-Programs 2Q161102B74FHovede Hall, West LaFayette, IN 47907 _____________

It. CONTROLLING OFFICE NAME AND ADDRESS M2 REPORT DATE .I

US Army Research Institute for the Behavioral and February 1984 ISocial Scien~ces 12. NUMBER OF PAGES5001 Eisenhower Ave. Alexandria, VA 22333 2

*14. MONITORING AGENCY NAME A ADDRESS(II diffeet barn ConU.Ubnd 018..) 15. SECURITY CLASS. (of Odle mrepa

UnclassifiedISo. DECLASSI FICATION/ DOWNGRADING

SCHEDULE

UL D8ISTSY40UTION STATEMENT (of Wet Re pmgt

Approved for public release; distribution unlimited

17. ISRIOUTION STATEMENT (of lbe "acat enterd in Bloak 2.it difeen he Repeal)

IS. SUPPLEMENTARY MOTES

19. KEY WORDS (Continue an revere.i aide ft neceeary and Identify by block num ber)

Absence behavior, job *satisfaction, pressu re variables, additive model, moder-ated model, regression moeel, intrinsic and extrinsic satisfaction, Minnesotaatisfaction Questionnaire (DISQ)

Tjio1 models of absence behaviors were compared. The first considered absencebehavior as a function of job satisfaction. Additional pressures towardattendance, both internal (the individual's value system) and external (co-workers' and job structure) as well as four demographic variables were consideredas moderators of the absenteeism-job satisfaction relationship. These additionalpressures were termed role pressures. The second model considered absenteeismas a function of role pressures and job vatisfaction in an additive rather than-

(continued)

DDI F OR 1473 ELIT-nilN OF* I NOV 6S IS OBSOLETE Ucasfe

SECURITY CLASMiFICATION OF TMIS PAGE (When Does Entered)

UnclassifiedSECURITY CLASSIFICATION OF THIS PAG0(Wh' Data Enateud)

b moderated fashion. Only the additive model was supported. The data werediscussed in light of the process of role pressures for attendance.o'This report is fourth in a 1975-1976 series entitled "Sources and Effectsof Accurate Work Perceptions."

b 5

.9

--

.. 9 ..,

-9. o -. *_ ( . - -.

." . .vv~b1t oe

a.;

Uenessoa ifri

r J tif I octiou ... ---

;. . Avail and/or

a..

,.TUnclassified: '$sCURoTY CL.ASSIFICATION OF' THS PAGCsbe- Date Eo,*,0.

L -W --

The Role of Job Satisfaction in Absence Behavior

From the organization's standpoint absenteeism is an important behavior.

Due to its importance there has been considerable interest in it. (See, for

example, reviews by Brayfield & Crockett, 1955; Gibson, 1966; Herzberg, Mausner,

* Peterson, & Capwell, 1957; Porter & Steers, 1973; and Vroom, 1964.)1 The

research is almost exclusively correlational, and the variables correlated

with absence behavior fall into two general classes. First, absences are

correlated with biographical, individual differences such as age (Cooper & Payne,

*1959), family size (Naylor & Vincent, 1959), tenure (Baumgartel & Sobol, 1959),

and sex (Isambert-Jamati, 1962). The second approach has been to correlate .

measures of job satisfaction with absence behavior. In general these studies

have related absenteeism to satisfaction with the job, organization-wide factors

(e.g., pay and promotion), imediate work enviornment factors (e.g., pupervision

and working conditions), and job content factors (e.g., Task repetitiveness,

autonomy, and responsibility). (See Porter & Steers, 1973, and Vroom, 1964, for

reviews focusing on job satisfaction).

Of the two general sources of absenteeism data - biographical and job

satisfaction - the latter assumes a process model which is seldom made explicit.

17h interest in absenteeism should not imply a consensus on what constitutes

absence behavior. For example, very different patterns or results have been

found when absences were defined as the frequency of absences than when defined

as the duration of absence (Metzner & Mann, 1953). At this point in our discussion

we wil use absences in the more generic sense to include all avoidable absences

from the Job; the exception would be those due to known illnesses. Later in

the article we will address specific types of absenteeism.

%a .. .- -

2

It assumes employees approach situations anticipated to lead to satisfaction

and avoid situations perceived to lead to dissatisfaction. It is reasoned

that the higher the satisfaction with the job the less the tendency to with-

draw temporarily from the job by being absent. In this case, attitudes about

the job are viewed as antecedents of behavior. This implies that, if the

employee's attitudes are known, his absence behavior can be predicted.

The research on the job satisfaction-absenteeism relationship is incon-

clusive; sometimes the tvo are correlated, other times they are not. (See Porter

and Steer's recent review, 1973). This lack;of a strong correlation between

job attitudes and absence behavior brings into question the general assumption

that temporary withdraw. from the job is caused by dissatisfaction with the job.

Two general theoretical explanations pertain to the low observed correlations

aside from the obvious one that the measures of job satisfaction and/or of

absenteeism have not been satisfactory. The first assumes that the model which

views attitudes as antecedents of behaviors is correct. This position argues

that the low correlation between attitudes and beha rlor is due to external

constraints which Influence behavior and do not allow the individual to behave

in line with his/her attitudes. Herman (1973) developed this position as it -

relates to the relationship between job satisfaction and performance. She argued

that the low correlation found between job satisfaction and behavior across a

large number of studies (cf., Brayfield 6 Crockett, 1955; Vroom, 1964) may have

been due to the fact that many investigators looked.at this relationship in

settings where the individuals had only limited control over their performance

behavior. An extreme example is the.asseubly line. Here an individual'k

performance Is almost exclusively a function of factors beyond his control and,

therefore, no relationship should be expected between performance and job

satisfaction. Hrman demonstrated that when external constraints on behavior

were removed by secret ballots in a union election, voting behavior was..

3

predictable from attitudes about unionization. In a similar view, Dansereau,

* Cashman and Green (1974) hypothesized that one of the constraints influencing

turnover was the individual's perception of the availability of suitable jobs.

Their data showed that for those who felt other jobs were likely to be available,

the correlation between job satisfaction and turnover was significantly higher

than for those who did not perceive alternative jobs to be as readily available.

For both Herman and Dansereau et al. a strong relationship between attitudes

and behaviors was assumed to exist and other factors influencing behavior were

conceptualized as moderator variables. If the influence of external constraints

on behavior is absent, job satisfaction should influence job behaviors; if

the constraints are present, the job satisfaction-behavior relationship is

decreased. In the case of absence behavior, the mote that absenteeism is

influenced by factors beyond the control of the individual the smaller should be

the job satisfaction-absenteeism relationship.

An alternative explanation for the low correlations observed between job

satisfaction and absenteeism behavior considers the decision to attend as a

function of several factors in addition to satisfaction with the job. Variables

asumed to be related to absence behavior would not be considered moderators

- of the job satisfaction-absenteeism relationship but would be considered

independent influences on attendance behavior.

.9 This model is most similar to that presented by Graen (1969) when he extended

the Expectancy Theory model of Vroom (1964) to include internal and external

role pressures. These pressures were assumed to add additional force toward

the behavior in question. With regards to absence behavior, this position

argues that satisfactton with the job provides onlyone force to attend. Other

independent forces orpressures are present in the job setting which do not

necessarily interact with satisfaction. This position assumes that a high

correlation between job satisfaction and absenteeism should not be expected

4

because, even under low constraints, the decision to attend on a daily basis

is only slightly related to one's job satisfaction.

With respect to other variables in addition to job satisfaction which

influence absenteeism behavior, the two explanations offered above employ

very different strategies. In the first case, other influences on absenteeism

behavior are viewed as constraints modifying the absenteeism-job satisfaction

relationship. It is assumed that if the constraints were removed absenteeism

behavior could ultimately be explained by job satisfaction. The second approach

considers the potential correlation between job satisfaction and absenteeism

as very limited in most work settings and views other variables as additional

influences of absenteeism behavior. In the latter case, such variables are

Independent predictors to be combined with job satisfaction in an additive

model, such as a regression model, in order to predict absenteeism. The purpose

of the present study was to compare these two models for the prediction of

absenteeism behavior.

Method

Procedure

Clerical workers in several departments at Purdue University participated

in the study. All clerical workers within the selected departments (see below

for the bases of departmental selection) wore informed of the general purpose

of the study, then were sent a questionnaire in April 1973. For those who

returned them, their attendance data were obtained from the personnel office for

the six month time period following the adminitration of the questionnaire.

Finally, those who had remained on the sam job and who had returned a question-

ivre were readministered the same questionnaire in November 1973 and again

in May 1974. The data reported here deal with the responses on the first

admnistration of the questionnaire and the attendance behavior from April to

November 1973 because high turnover made successive samples unrepresentative of

the population.

Measures

Job satisfaction. Job satisfaction was measured by the complete Minnesota

Satisfaction Questionnaire (MSQ) (Weiss, Dawis, England, & Lofquist, 1967).

The scale contains twenty subscales each with five items. The scale was used

in order to measure satisfaction with several intrinsic and several extrinsic

job factors. Intrinsic job factors are those which are provided to the indivi-

dual by some agent (e.g., co-workers, supervisors, or the organization) in the

work setting.

The booklet for scoring the MSQ gives an intrinsic and an extriuasic sub-

total score. However, the subscales which comprise the two dimensions include

several job factors that do not fit the more recent distinction between intrinsic

and extrinsic job facets. For example, authority is considered an intrinsic

factor. Although this is consistent with Herzberg's classification (Herzberg,

Hausner, & Syndermen, 1959) it is not consistent with the more recent distiction

between intrinsic and extrinsic factors. Therefore, the following subscales

were used to define intrinsic satisfaction: ability utilization, achievement,

creativity, and responsibility. Extrinsic satisfaction was defined by: company

policy and practices, compensation, co-workers, security, social status, super-

vision-- human relations, supervision - technical, working conditions. Total

job satisfaction was the sum of the twenty subscale scores of the MSQ.

Test-retest reliabilities for the three satisfaction scores over six month

and one year per16ds are shown in Table 1. Given the length of the time period,

these reliabilities were considered very satisfactory. The correlation of

Intrinsic satisfaction with Extrinsic satisfaction on the April 1973 administration

of the scale was U- .64, N - 165)'indicating that the subscales shared a

considerable amount of common variance.

y Role pressures. Role pressures were indexed by two sets of measures.

First biographical data were collected as indices of role pressures on the

6

Table 1

Test-Retest Reliabilities of Satisfaction and Pressure Measures

Correlations of Correlations of Correlations ofApril 1973 with November 1973 with April 1973 with

November 1973 Measures May 1974 Measures May 1974 Mfeasure-

(N -" 100) (N -" 82) (N -" 82)

Satisfaction

KSQ Intrinsic .66a .69 .60

MSQ Extrinsic .61 .65 .56

MSQ Total .67 .65 .62

Pressure

Value System .68 .70 .69

Co-worker .48 .58 .53

Job Structure .63 .67 .64

Total Pressure .62 .66 .72

afor all r's, p< .001

9'- . , -- % -_.o, - - ,,. - - . , . ,-,-,-% , -,--"%-% ,- ' , "%,, '%"% ,o . . , . -. .

.7

individuals due to such off-the-Job factors as their family responsibilities.

The second set of pressure measures were based upon job and value system

pressures the participants felt in the work setting. Each set is described

below.

4 2 Biographical measures: Four biographical measures were obtained. These

* were age, number of children living at home, whether or not these children were

under seven years old, and, if married, whether or not their husband was a

student. The last item was included because the university employs a large -

number of student's wives and it was thought that academic schedules which

create several time periods during the year when students are on vacation might

influence absence behavior of students' wives.

Perceived pressure variables: Three variables were designed to measure

several sources of pressure on the women to attend-work not reflected by bio-

graphical variables. With one exception, all items were measured with five-

point Likert scales ranging from strongly disagree to strongly agree. The

first of these dealt with'-the pressure for attendance created by the Job

structure. Job Structure Pressure was defined by two items. The first asked

the degree to which other clerical workers in the sae department completed

the individual's work when she was absent. It was reasoned that the more that

work piled up on the job during her absence, the more pressure there should

be for attendance in order to avoid having to work harder the first few days

back on the job. This job pressure item was:

a. If I am absent, I have to work harder the first day or so when I get

back due to the work that piled up in my absence.

The second job pressure item described five types of job structures for

clerical jobs in this setting. These descriptions were based upon interviews

with clerical staff and are listed below:

A. Individual secretaries are assigned to specific professors, individuals,

8

or specific functions. They work only on work from those persons or areas.

If a secretary is absent the work piles up. Only work that has to get done

i=ediately would be given to someone else in her absence.

B. Individual secretaries are loosely assigned to specific professors,i individuals, or projects. However, work is often distributed without regard

for these assignments. Often, when one person has a lot of work, some of this

is distributed to others who are less busy.

r C. Secretaries generally-operate as a pool. However, members of the pooi

are often assigned to specific projects for a few days or weeks ,nL during this

time are pulled out of the pool.

] D. Secretaries operate as a pool. New work loads are dis .. ted to whom-

ever is least busy.

E. Secretaries are assigned specific professors, individuals, or specific

• functions in teams of two or three. Each secretary within the team works on

specific projects, but if she is behind or is absent the other team members do

her work when they have the time.

The respondents chose the one response which best described their job.

For scoring purposes the item was coded in the following manner: A - 5; B, C,

or E - 3; and D - 1. This score was then added to the one Likert item above.

A second pressure variable was called Value System Pressure. This was

designed to measure the extent to which the individual believed that absence

was wrong. Value pressure was measured by the following items:

e. I believe that, once you accept a job, you are obligated to go to

work unless you absolutely cannot make it.

b. As long as I have sick leave days availkble, I see nothing wrong with

using them as I wish. (reverse scored)

c. When someone is absent without a good reason, I dislike doing their

work for them.

! ~ *. '*., - ,, . . . .. ,. ,..,., . . . .... . *.&*.. -. ..

9

A final pressure variable measured the extent to which co-workers

'U accepted absence behavior. Three items defined Co-workers Pressure:

a. Even on days when I would rather stay home, I come to work because

I feel a strong loyalty to the people I work for.

b. On days I would rather stay home, I come to work because I know if I am

not here the other secretaries with whom I work will have to work harder.

c. Many people with whom I work use sick leave days for thdir own convenience

rather than for illness. (reverse scored)

d. Most People I work with do not mind filling in for someone who is absent

for any reason. (reverse scored)

A final pressure score, Total Pressure, was created by adding the pressure

-- . variables. Table 1 gives the test-retest reliabilities for each of the pressure

-,

Svariables.

Sample

The sample consisted of clerical workers who u_ rked In thirteen different

academic or administrative departments of the university. The departments

were selected at random with one exception. An attempt was made to select

departments that varied along the dimension of job structure for their clerical

staff. In order to investigate the effect of job structure 9ressure, the sampling

attempted to have departments that used a secretarial pool (low job structure

pressure) as well as the more typical structure in which one secretary worked

U" for several professors. Therefore, two departments known to have secretariil

pools were included and not selected randomly. For those departments or

administrative units included in the sample, all full-time secretaries in the

department, not in supervisory positions, were included. The selected

departments employed from 4 to 35 clerical workers. Two hundred forty-three

questionnaires were mailed out and one hundred seventy-seven were returned

- . . .. . . ... .

. .10

which represented seventy-three percent of the sample.

Absenteeism

*Absenteeism was defined by the total time absent in the period from

28 April 1973 to 13 November 1973. From the university personnel records

it was possible to break this down into the number of hours of sick leave

used during the time period and the number of hours of unexcused absences.

These hours were converted to a ratio of number of hours absent over total

number of hours possible to work because many (52) of our sample quit before

the end of this six month time period. For:-turnovers, the divisor was the

number of possible hours for work from the 28th of April to the day they quit.%..

Three absence measures were used: amount of sick leave, unexcused absences,

and total-absences. Unexcused absences were all absences for which the individual

did not use sick leave. No attempt was made to judge whether the absence was

really unexcused. To consider all absences other than sick leave unexcused

was a conservative strategy which may have decreased the strength of the results.

However, we did not want to rely on reported reasons of unexcused absences in

order to make some decision on the legitimacy of the absence. In most work

settings, to implement company policies about what is and is not an unexcused

absence usually yields some complex mixture of the integrity and the creativity

of the work force rather than an accurate measure of truly unexcused absences.

All absence measures represented only measures of duration rather than

the frequency of absence. Some data exist which show that frequency is a

better measure than duration (Metzer & Mann, 1953). Unfortunately, it was

not possible to obtain frequency data from the records available in the study.

However, to remove the bias in the sick leave measure created by those who

had been absent for long periods of time, it was decided to eliminate from the

sample all those whose length of absence for sickness was greater than +3

V.

standard deviations above the me. The cut-off was based on an inspection

of the sick leave data which showed 11 individuals who fell well outside the

distribution of the rest of the sample. Therefore, the primary data analysesS%.. .

were based upon a sample of 166 of which missing data on specific scales reduced

the number for some of the correlations presented.

Results

Zero-Order Correlations

Table 2 presents the means and standard deviations of the satisfaction,

role pressure, and absence measures, and Table 3 presents the intercorrelations

of these measure. Several of these correlations are of interest.

First, none of the satisfaction measures correlated significantly with

any of the absence behaviors. Although significant correlations have been

found in previous research, the lack of any significant zero-order correlations

is not inconsistent with either the moderated or the additive models considered

in this study.

Three of the four biographical variables correlated significantly with

one or more of the absence measures. Older women used less unexcused absences and

were absent less overall than were younger women although the latter correlation

was due primarily to the contribution of unexcused absences to the total

amount of absenteeism.

Women whose husbands were students used significantly more sick leave than

those whose husbands were notandrtheir total number of absences also were higher.

It was predicted that this subgroup would be more likely to be absent; it was

not predicted that they would choose sick leave rather than unexcused absences

w en not attending work. However, since this subgroup was a transient one

comprised of individuals who typically live in the community only for the time

12

Table 2

Means and Standard Deviations of Job Satisfaction, Pressure,

Biographical, and Absenteeism Measures

Variables Mean Standard Deviation N

Job Satisfaction

MSQ Intrinsic 75.02 14.68 165

MSQ Extrinsic 141.64 23.71 165

MSQ Total 366.48 56.02 164

Biographical Variables

Age (1-16-24; 2-25-29;3-30-39; 4-40-49;5-> 49) 2.31 1.53 165

Husband a student? (1-yes;

2-no) 1.18 0.83 165

Number of Children .61 1.11 165

Number of children under7 years old .15 0.35 165

Pressure Variable

Value System 11.08 2.10 165

Co-workers 13.09 2.07 165

Job Structure 8.14 1.82 154

Total 32.32 4.16 153

Absenteeism

Sick Leave 0.031 0.017 166

'Unexcused Absences 0:013 0.035 141

Total Absences 0.045 0.038 141

i

4I -• . . ,"- ¢ ...i,*1 .. . ... ,.. ,.,- , -'. .,' . . , ,

-0-W-jM - I -n -XI-M -An. -,- C. WT W TV 9

13

Table 3

Intercorrelations of Job Satisfaction, Biographical

Pressure and Absenteeism Measuresa,b

1 2 3 4 5 6 7 8 9 10 11 12 13tisfaction 045Q)a1. Intrinsic 64e 89e 29e -09 09 -02 28e 29e -01 26e 01 -102.- Extrinsic 89e 27e 06 16c -01 21d 28e -06 21d 05 -10S3. Total 33 e 09 15c -02 27e 35e -03 28

e 03 -09

Slographical Data

4. Age 2 7e

2 7e -10 4 3e 30e 21d 44

e -06 -20d

5. Husband Student? 38e 12 09 11 03 13 -26e -076. Number of Children at Home 14 23d 13 06 22d -10 -027. Children Under 7 years.old 10 10 09 16c 09 15c

assure Variables

8. Value System 45e 17c 80• _15c 079. Co-worker 03 74e -02 20

10. Job Structure 5 4 e - 2 0d -1111. Total -15 10

Asenteeism

12;. Sick Leave -0413. Unexcused14. Total

-a - 165 except for correlations with the following:N - 164 for KSQ total

N N - 154 for Job Structure Pressure

N - 153 for Total Job PressureN - 141 for Unexcused and Total Absences

b Decimals omitted.~c

< .05

d

<.001

14

their husbands are in.school, it is quite possible that they felt little need

to accumulate sick leave for possible use in the distant future and/or they

were less committed to the organization and more willing to use sick leave

regardless of the actual reason Tor absence. Both interpretations are cone

sistent with the observed correlations.

The third biographical variable of interest to absenteeism is family

responsibility -- specifically the number of children. Only the number of

children under seven significantly correlated with any form of absence behavior.

Those who had children under seven used more unexcused absences than did those

who did not have children in this age range. As was predicted, children only

affect absence behavior when they are at the age which demands adult care.

Turing to the perceived pressure variables measured, all were directly

related to some form of absence behavior. First, the internal pressure from

their value system did influence the use of sick leave. The greater the

pressure, the less they used sick leave. Second, there was a positive

correlation between co-worker pressure and unexcused absences. The direction

of this correlation becomes interpretable upon closer inspection of the measure

of co-worker pressure. One of the items deals only with pressure not to use

sick leave, therefore, the more the women perceived a norm for using sick leave

only when they were actually sick, the more they tended to use unexcused absences

when they wanted to miss work. The perception of co-worker pressure did not

Influence the total number of absences, but it did influence how the absence

behavior was manifested.

Job structure pressure, as predicted,nalso influenced absenteeism. There

were fever total hours absent and fewer hours of sick leave used by those who

felt their job demanded their presence.

M derated vs. Additive Prediction

The major focus of the study was to compare pressure variables as moderators

15

of the satisfaction-absenteeism relationship to their use in an additive

model for the prediction of absenteeism. For the moderated prediction, the

sample was split at the median on each of the following measures: Value

". System Pressure, Co-worker Pressure, Job Structure Pressure, Total Pressure,

Age, and number of children living at home. In addition, the sample was

divided once for each of two dichotomous variables -- whether or not their

husband was a student, and whether or not they had children under seven years

old. For all variables it was predicted that the correlations between job

satisfaction and absenteeism would be lower in magnitude when the pressures

were high than when they were low.

For each factor mentioned above, nine correlations were calculated for

those above the median (or, in the case of the two dichotomous variables,

within the high category) and nine for below. These nine were intercorrelations

of intrinsic, extrinsic, and total s8tidfaction with sick leave, unexcused

. *absences, and total absences. The median correlations under low and high

pressure conditions are reported in Table 4. The correlations on which the

medians were based ranged from -.21 to +.15 under low pressure and from -.20

to +.29 under high. Only one of the eight moderator variables showed the

predicted pattern of correlations, and this only occurred for two of the

three absence measures. The job satisfaction of those who did not have

children under seven years old significantly predicted unexcused and total

absences. The correlations with unexcused absences were -.19, -.19, and -.20

for Intrinsic, Extrinsic, and Total Job Satisfaction, respectively; those with

Total Absences were -.19, -.15, and -.19. All were significant at the p < .05

level. In addition, all these corcelations were significantly different

from their matched correlation in the high pressure condition (having children

under seven years old). None of the correlations in the latter group were

s gificant.

-- ) -%a -- Ir q -

16

In spite of the moderated effect for children under seven, overall, little

support was found for the moderated effect. Only one of eight moderators was

successful, and this one was successful for only two of the absence measures.

Furthermore, those correlations observed under the low pressure condition,

although significant, were quite low.

To investigate the additive model for comparison to the moderated one,

multiple regression analyses were conducted. Table 5 presents these data for

both types of role pressures separately and in combination. In general, the

data support the additive interpretation. When knowledge of all forms of role

pressure assessed in the study was added to job satisfaction, it was possible

to predict all forms of absenteeism behavior, and these predictions held up

when shrinkage was taken into account. Within types of role pressure, only

perceived role pressures reliably predicted absenteeism, in this case Total

Absences. None of the multiple correlations with biographical variables held

up after shrinkage was taken into account.

Discussion

In the present sample, the conclusion must be reached that job satisfaction

was, for the most part, unrelated to absenteeism. This was true when satisfaction

was related directly to absenteeism and when the possibility of a moderated

relationship was explored. Three competing explanations for the lack of relation-

* ship exist: the appropriate moderators were not measured, the measures were

not sufficiently reliable or valid to detect a difference, or job satisfaction

had very little relationship to absenteeism.

Whether or not the appropriate moderators were measured cannot be answered

from the data. However, the moderators used were not selected haphazardly.

Th. variables were chosen only after interviews with clerical staff and their

supervisors and after critical evaluation of the role pressures likely to be

present on the job. Therefore, it seem unlikely that all of the moderators

were inappropriate.

7. 7% .n 777

17

Table 4

* Median Correlations Between Job Satisfaction and Absenteeism under

-- High and Low External Constraints as Measured by Pressure and Biographical Variables

Unexcused TotalSick Leave Absences Absences

Job Satisfaction (MSQ) Lo; Levels of Constraints

Intrinsic .01 -.08 -.11

Extrinsic .07 -.15 -.11

Total .06 -.14 -.11

High Levels of Constraints

Intrinsic .02 -.06 -.08

Extrinsic -.01 -.03 -.05

Total .00 -.03 -O

W .- S ... -A-- ** .

LO 01 5-8 C % AN 5 C

IA Ah CA C %-a

W-0 0t -0 0 O M I0 W

ca n wD a0 0 W 0 0At "5 "1 0- 0& 'is Mi

f %0.ft l n .m. a CL to Cj L. 10 1. -i " *m* m ~ 1 t

mnW 4 1~ to0Q C S S f

0% %D'4 0L Ol wt w 5 %4 nt a -' -t c 0

s-es 2.0 0' zCft

a I'& P& 6t '1 n "M sn ' n ft 0.0~~~~~ C n S 0m m m

fi0 1 C . w0 lb IC FA c D

w A ~ A 'U) AU fO f? f t C0-w 0b at ft 5 5 S t

%. f t % -a *.a -C a a mft 4. %40~ 0I a. ral

*c a 40 pffS 'I0 '1~ 0 sa pfft

CP co -L% aa.a

0 n 0 0 a I 0

CO ONl Wo 8.8. wa w 0 a

0 a

V :01W~~~~~~ a d s'K 5 0 C

'0 'P US 0'0 w n g~iOva

01 01 1 OOC 0 5 f

19

Several factors militate against the second alternative that the variables

were insufficiently measured. First of all, the job satisfaction measure, the

MSQ, has received extensive use and has been found to be reliable and valid

.i- (Weiss, Dawis, England, & Lofquist, 1967). Although in the present sample,

test-retest reliabilities were only moderately high (see Table 1), it should be

recalled that the time period between test administrations was considerably longer

than is customary for test-retest reliability-estimates. Furthermore, when

the time period was extended to one year, the reliabilities still held up.

4. With regard to the absence behavior measures, it was pointed out earlier

that Metzer and Mann (1953) found frequency of absence to be a more appropriate

measure than duration. Although we were unable to measure frequency, it was

felt that the removal of the eleven clerical workers with extended periods of

sick leave (i.e., greater than +3.0 standard deviations above the mean on

sick leave) should have left a duration measure which was highly correlated

with frequency. Unfortunately, no adequate test of this supposition was

available,

The strongest support for the adequacy of the absence behavior is its

predictability from role pressure variables. Sick leave, unexcused absences,

and total absences all were related in a theoretically meanitgful fashion

to one or more of the perceived pressure or biographical measures. Therefore,

- it cannot be argued that the absence measures were in~alid.

We are left with the conclusion that job satisfaction was not very strongly

associated with the absence behaftor in this setting and may not be in many

other settings. Although such a condlusion lacks the appeal of more contemporary

values about the nature of work such as those in the R.E.W. report, Work

k mrica (1973), the conclusion is not out of line with the data. Vroom

(1964) reviewed ten studies relating job satisfaction to absenteeism and found a

"'IN

20

range in correlations of only moderate strength. Porter and Steers (1973)

updated this review and found very little work done in the last few years.

Although they concluded that satisfaction was related to absenteeism, Porter and

Steers tempered this conclusion by stating it was based on "preliminary evider.ce"

(p. 167). When one considers that the reveiws are based irimarily upon

* published articles and one takes into account the bias against publishing negative

findings, the confidence in job satisfaction-absenteeism relationship is shaken

even more.

In the face of unsupporting data, the tendency has been to question the

validity of the absence measures or to assume that other factors controlled behavior

so that the attitudes toward the job did not have a chance to inflilence it.

However, the present study was unable to demonstrate a relationship between

absenteeism and satisfaction when several of the possible outside factors

. vere removed. Thus, it is time that the assumed relationship between job

satisfaction and absenteeism be questioned.

Two general views have been espoused about the causal relationship between

job satisfaction and behavior. The first assumes that satisfaction causes

the behavior; the second assumes that the behavior leads to satisfaction

(Lavler & Porter, 1967; Pritchard, Kirk, & Mayo, 1975). Absenteeism usually

has been assumed to follow the first model. This model stems from the social

psychological literature which views attitudes as leading to an intention of

an approach or avoidance behavior toward the attitude object (Triandis, 1971).

With respect to absenteeism, this model assumes the greater the job satisfaction

the greater the display of approach behavior toward the job in the form of

attendance assuming thatt external constraints on attendance behavior are low.

The moderated correlations of the present study failed to support the

satisfaction-to-absenteeism model. Yet, a closer look at many of the job

* . , ~ V.- ~ ... , *. *J.-..%

21

dimensions typically measured by job satisfaction scales leads one to question

how attendance could ever be construed as an approach behavior resulting from

satisfaction with these dimensions. For example, it is unlikely that satisfaction

-- with pay, security, the company policy would lead to a greater desire for very

regular attendance in organizations with liberal sick leave benefits or with

seniority-based decisions about layoffs, promotions, and raises. In fact,

company policies which allow for more frequent absences may provide a basis for

greater satisfaction but certainly would not be expected to lead to greater

attendance. Baum (personal conuunication) has found a positive correlation

between absenteeism and the extent of overtime available. To the extent that

availability of overtime increases satisfaction with pay, this would imply a

positive correlation between pay satisfaction and attendance -- a prediction

opposite of the satisfaction-to-behavior model. Likewise, if there is no

4. reason to expect one's position with one's co-workers to be jeopardized by

occasional absences, satisfaction with co-workers should not necessarily pre-

*4 vent occasional absences. It is hard to imagine the attractiveness of any work

group being so strong that, if it administered no negative sanctions, its members

would still desire to always be present for work. Only satisfaction with

supervision appears to create a force toward attendance and, more than likely,

this force is due primarily to the power the supervisor holds over the indivi-

dual's attainment of valued rewards.

Theoretically intrinsic satisfaction with the work itself should create

more of an intention to attend work. However, even under conditions of

intense involvement with the job, absence behavior may occur. Professorial

positions in university settings offer a good example of a job filled by

members who possess relatively high intrinsic satisfactions and who experience

very few sanctions for not being in the office or lab from 8:00 to 5:00 every

22

day. Yet, it is a rare professor, indeed, who does not occasionally wake up

in the morning and consider and pursue other activities which are more attrative

than going to the office or laboratory, in spite of the fact that he may be

very satisfied with his work. Therefore, even for intrinsic job satisfaction,

the assumption that absenteeism should be linearly related to job satisfaction

appears to have been somewhat naive. &' model that assumes that absence behavior

results from the intended behavior of approaching the job due to the individual's

satisfaction with it does not seem adequate for the explanation of absence behavior.

The second model assumes that attendance behavior leads to satisfaction

to the extent that rewards are contingent upon attendance (Lawler & Porter,

1967; Porter & Lawler, 1968; Vroom, 1964). given this model, the explanation

for low correlations between job satisfaction and absenteeism is very straight-

forward; in most organizations very few rewards or sanctions are perceived

to be tied to attendance behaivors. Sick leave, personal lbave pblicies, as

vell as a decrease in a general work ethic, that it is one's duty to attend

* the job one holds, all tend to decrease the contingency between behavior and

rewards. Given the recent trends in company pblicies and practices, it may

become even more difficult to find job satisfaction-absenteeism correlations

4' in the future. Furthermore, we would predict that those who have found job

satisfaction-absenteeism correlations in the past have done so primarily

because the job environments were those in which rewards were contingent upon

attendance. Since such contingencies- :usually aren't very stroig, it. is. not

surprising that the correlations of job satisfaction with absenteeism have

been low.

The lack of support for the position that job satisfaction ckuses attendance

does not imply that absence behai-ior is any less important or less researchable.

.%%

4"

-~~~~~~~ -- iL IL* - -*~-~~~-I--p*. * . .. ' '-T.- 7W P "6 NV" . -

23

It does imply that a more efficacious approach to studying it must be

undertaken than to merely correlate it with satisfaction measures. The

data presented here suggest that attendance behavior should be viewed in

light of the forces provided by role senders who may or may not be present

* in the job setting. These roles may be sent by the organization through

company policies, the nature of the job, supervisors or co-workers, or by

agents less under the control of the organization -- the individual's work

values, his responsibilities, family, or warm spring days and nearby salmon

running. The present data have shown that role pressures provided by the

individual's value system, co-workers, and the job design do influence one

or more forms of absence behavior. What is heeded is a more comprhensive

view of the sent roles as they relate to attendance behavior (Gibson, 1966;

Hyman, 1955). Such an orientation would explore the link between agents as

* they influence the individual's perception of the role demands on his attendance

4 behavior created by the rewards administered by these agents. As being

* present at work becomes more instrumental for the attainment of valued rewards,

absenteeism should decrease. The focus of an organization interested in

absenteeism behavior must be on those agents within the organization who can

provide valued rewards for attendance behaviors. The concern with job

satisfaction is only an indirect one. Whether or not job satisfaction corre-

lates with absenteeism only depends upon whether they share a common third

factor -- the attainment of valued rewards.

N

'4

.,,,.- ,-.... ,,.-, ... ,,- ... ,. •..... ..... ,-........-... -... "... ... ,..,', . ....... ., .. . , -. *%,~ ~ b ,.,,. ,,,'.

References

Baumgartel, H., & Sobol, R. Background and organizational factors in absenteeism.

Personnel Psychology, 1959, 12, 431-443.

Brayfield, A. H., & Crockett, W. H. Employee attitudes and employee performance.

Psychological Bulletin, 1955, 52, 396-424.

Burket, C. A study of reduced rank model for multiple prediction. Richmond,

Virginia: William Byrd, 1964.

Cooper, R., & Payne, R. Age and absence: A longitudinal study in three firms.

Occupational Psychology, 1965, 39, 31-43.

Dansereau, F., Cashmen, S., & Graen, C. Expectancy as a moderator of the relation-

ship between job attitudes and turnover. Journal of Applied Psychology,

1974, 59, 228-229.

Gibson, R. 0. Toward a conceptualization of absence behavior of personnel in

organizations. Administrative Science Quarterly, 1966, 11, 107-133.

Graen, G. Instrumentality theory of work motivation: Some experimental results

and suggested modifications. Journal of Applied Psychology, Monograph,

1969, 53(2), Part 2, 1-25.

Health, Education, and Welfare. Work in America. Cambridge, Massachusetts:

MIT Press, 1973.

Herman, J. B. Are situational contingencies limiting job attitude - job

- ,performance relationships? Organizational Behavior and Human Performance,

1973, 10, 208-224.

Herzberg, F., Mausner, B., Peterson, R. 0., & Capwell, D. F. Job attitudes:

Review of Research and opinion. Pittsburgh: Psychological Services of

Pittsburgh, 1957.

Herzberg, F., Mausner, B., & Synderman, B. The motivation to work. New York:

Wiley, 1959.

. .I-.... . _. . - .. . . ..-

Hyman, H. Survey design and analysis. Glencoe, Illin6is: The Free Press,

1955.

Isambert-Jamati, V. Absenteeism among women workers in industry. International

St. Labor Review, 1962, 85, 248-261.

Lawler, E. E. III, & Porter, L. W. The effect of performance on job satisfaction.

Industrial Relations, 1967, 7, 20-28.

Metzner, H., & Mann, F. Employee attitudes and absences. Personnel Psychology,

1953, 6., 467-485.

Naylor, J. C., & Vincent, N. L. Predicting female absenteeism. Personnel

Psychology, 1959, 12, 81-84.

Porter, L. W., & Lawler, E. E. III. Managerial attitudes and performance.

Homewood, Illinois: Irwin, 1968.

Poter, L. W., & Steers, R. W. Organizational, work, and personal factors in

employee turnover and absenteeism. Psychological Bulletin, 1973, 80,

151-176.

Pritchard, R. D., Kirk, R., & Mayo, R. Some data and comments on the relation-

ship between performance and satisfaction. Lafayette, Indiana: Purdue

University, unpublished paper, 1975.

Triandis, H. T. Attitude and attitude change. New York: Wiley, 1971.

Vroom, V. H. Work and motivation. New York: Wiley, 1964.

* Weiss, D., Davis, R., England, C., & Lofquist, L. Manual for Minnesota

Satisfaction Questionnaire. Minneapolis: University of Minnesota,

Minnesota Studies in Vocational Rehabilitation, 1967, XXII.

°